Considerable evidence has now accumulated demonstrating the efficacy of Individual Child Cognitive Behavior Therapy for reducing anxiety disorders in children. In growing recognition that the child's context affects the development, course, and outcome of childhood psychopathology and functional status, recent clinical research efforts have been directed toward evaluating whether cognitive behavior therapy when used with anxious children also is efficacious when particular contexts (i.e., family/parents, group/peer) are incorporated in the treatment program. As a result, there now exists considerable empirical evidence that childhood anxiety disorders also can be reduced in cognitive behavioral treatment programs that incorporate family/parents and peer/group contexts and target specific domains/content areas relevant to these contexts. Despite the above, there have been no studies that have directly evaluated whether family/parents and peer/group interventions that target specific domains/variables and content areas relevant to that respective intervention context actually produce specific effects on these domains/variables and, more importantly, whether changes produced on these variables mediate treatment response. Consequently, claims regarding the importance of incorporating (or not incorporating) family/parents and peer/groups and targeting respective variables relevant to each context in order to produce child treatment response are based more on speculation than on empirical data. Investigating whether incorporating family/parents or peer/group contexts and targeting specific domains/variables and content areas relevant to these respective contexts, and whether changes on these variables mediate treatment response in two cognitive behavioral treatments that each represent these distinct contexts (i.e., family/parents and peer/group) among children with anxiety disorders, thus comprise the specific aims of this project. The study targets the same DSM-IV anxiety disorders targeted in previous clinical trials and that are most common in children: social phobia, generalized anxiety disorder, and separation anxiety disorder. Using a controlled clinical trial design, 252 children (ages 8-14 years) and their parents will be admitted to treatment over the five years of the study, yielding an estimated 216 treatment completers at post-test and 180 at one year follow- up. Children and their parents will be randomly assigned to one of two treatment conditions: Family/Parents Cognitive Behavior Therapy (FCBT) and Peer/Group Cognitive Behavior Therapy (GCBT). All participants will be assessed at pretreatment, posttest, and one-year followup. Two sets of hypotheses will be tested. Because each condition represents a distinct treatment context (family/parents and peer/group) that targets the same two domains (skills and relationships) but in two different content areas within each domain (i.e., parenting skills and parent-child relationships in FCBT versus child social skills and peer-child relationships in GCBT), the first set of hypotheses is designed to establish empirically whether there are in fact treatment specific effects. Thus, the first set of hypotheses to be tested is that FCBT will produce significantly greater specific effects on parenting skills and parent-child relationships than on child social skills and peer-child relationships. GCBT, on the other hand, will produce significantly greater specific effects on child social skills and peer-child relationships than on parenting skills and parent-child relationships. The second and more theoretically and practically significant set of hypotheses will test whether or not it is the changes that are produced on these variables that mediate treatment response. Thus, the second set of hypotheses to be tested is that parenting skills, parent-child relationships, child social skills and/or peer-child relationships will be significant mediators of treatment response, i.e., anxiety reduction. To test the study's mediational models and to fully examine specificity effects, a multi-analytic approach that includes structural equational modeling and other complex data analytic strategies (e.g., growth curve modeling) will be used.